开关电容电路结构的神经网络优化设计

Optimization Design by Neural Networks of Switch Capacity Circuit Structures

基于神经网络解决优化组合问题原理，将计算能量函数与罚函数有机地进行了统一。构造了期望的罚函数，因而设计了结构非常简单的开关电容电路，并讨论了其特点，给出了其寻优步骤。计算机仿真结果表明，无论起点在何处，均能找到理论上的优化值；另外，目标函数的一些参数随时间随意变化时，变量可实时优化。

Based on the theory of solving optimization combination problems by using neural networks, the computational energy functions are united with the penalty functions. The expective penalty functions are constructed, so switched capacity circuits with simple structures are designed, their properties are investigated, and the procedure of seeking optimizationis given.The results by computer simulations illustrate that wherever the starting points are, the theoretical optimal value can be obtained; moreover, when some parameters of objective functions arbitrarily vary with time, variables can realize real time optimization.